|Publication number||US8064510 B1|
|Application number||US 11/724,881|
|Publication date||Nov 22, 2011|
|Filing date||Mar 15, 2007|
|Priority date||Mar 15, 2007|
|Publication number||11724881, 724881, US 8064510 B1, US 8064510B1, US-B1-8064510, US8064510 B1, US8064510B1|
|Inventors||Joseph N. Babanezhad|
|Original Assignee||Netlogic Microsystems, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (37), Non-Patent Citations (12), Referenced by (4), Classifications (11), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application relates to commonly assigned U.S. patent application Ser. No. 11/724,817, entitled “New Least Mean Square (LMS) Engine for Multilevel Signal” filed on Mar. 15, 2007 by Joseph N. Babanezhad, U.S. patent application Ser. No. 11/724,816 entitled “Joint Phased Training of Equalizer and Echo Canceller” filed on Mar. 15, 2007 by Joseph N. Babanezhad, the disclosures of which are hereby expressly incorporated herein by reference.
The invention relates to the field of analog signal processing.
Digital signal processing is widely used to process data carrying signals to remove, for example, inter-symbol interference (ISI), echoes, cross talk and other impairments, and to provide filtering, correlation and other signal processing functions. Today, numerous analog signals are processed in the digital domain. Typically, after some analog filtering and amplification, the analog signal is converted to a digital signal for digital signal processing. The design of the analog-to-digital (A/D) converter can become critical, particularly, as baud rates increase. In fact, in some applications, the design of an A/D converter may be considered a limiting factor.
The digital signal processing section may include a feed forward equalizer (FFE) 19 to remove such impairments as precursor ISI caused by insertion loss. The output of equalizer 19 is summed with an output from feedback equalizer 27 for past cursor ISI. The outputs of echo and cross talk cancellers can also be summed at unit 26. The output of summation unit 26 is the digital output signal provided on line 20. An error signal, which is generated by slicer 21, is used by the digital signal processing section. The input and output of slicer 21 are subtracted from one another by subtractor 24 to provide the error signal on line 25. This error signal is coupled to both equalizers 19 and 27. The output of summation unit 26 is also input to slicer 21.
The signal-to-noise ratio for the arrangement of
Conventional Phased Training of AEQ and AEC
A conventional transceiver that includes an equalizer (AEQ) and an echo canceller (AEC) is trained using conventional phased training. However, in the conventional phased training, the training of the AEC and the AEQ are performed sequentially in either the digital or analog domains.
In the digital domain, the AEQ and the AEC are trained separately in digital signal processing. In digital signal processing, when determining the proper tap weight, the proper tap weight is set or stored, and then after training both the AEQ and the AEC separately, they are run together in what is called “show-time” (also known as normal mode) using the stored tap weights. In other words, the AEQ is trained first, resulting in the tap weights for the equalizer being obtained and stored. Then, after the training of the AEQ has finished, the AEC is trained, resulting in the tap weights for the echo canceller being obtained and stored. This conventional training process requires a complex start-up process and complex timing, as described below with respect to phased training in the analog domain.
Training the AEQ and the AEC, in the analog domain, presents an issue that the tap weight can not be stored or set indefinitely, as done in phased training in the digital domain, but only temporarily for a short time. Thus, training must continue until the AEQ and the AEC are suitably trained.
Phased training the AEQ and the AEC in the analog domain, generally is performed by training the AEQ, storing the tap weights of the AEQ, and then subsequently training the AEC. The tap weights of the AEQ are stored digitally. Phased training in the analog domain requires a complex start-up process. The start-up process requires that while training the AEQ, the operations of the AEC be halted. Then when subsequently training the AEC, the operations of the AEQ are halted. This requires complex timing to halt and starting the operations of the AEC and AEQ.
Using this approach, the performance of both the AEC and AEQ depends on each other because phased training is based on a shared error signal. For example, the error signal that is used for the AEQ is also used for the AEC. For example, when the AEQ has not trained properly (e.g., shared error signal is large, such that the signal-to-noise ratio (SNR) is less than 10 dB), the subsequent training of the AEC will be impacted, and consequently, will not train properly as well. The large error signal may be the result of having an incorrect tap weight, and then by subsequently using that error signal, there may be too much echo, which consequently, yields unstable loops in the system. In conventional phased training, both the AEQ and AEC may each have a corresponding least mean square (LMS) machine. The LMS machines each look at an error signal, which is shared between the two LMS machines. The LMS machine of the AEQ receives an error signal that is generated from the output signal of a subtractor that is coupled to the output of the slicer. The subtractor receives the output signal of slicer and subtracts the output signal from AEQ. The input to LMS machine of the AEC is coupled to the output of same subtractor that is coupled to the output of the slicer.
It should be noted that although slicer 80 is illustrated as including A/D converter 81 and D/A converter 82 in the same component, A/D converter 81 and D/A converter 82 may reside on separate components. For example, A/D converter 81 may be A/D converter 18, as described with respect to
The conventional slicer introduces unwanted noise and unwanted spikes in the error signal. Because slicer 80 includes A/D converter 81, slicer 80 introduces quantization noise into the system, which affects the signal-to-noise ratio. As described above, the signal-to-noise ratio is a function of the near end and far end alien crosstalk, line noise, uncancelled impairments that result from factors such as line insertion loss, return loss, crosstalk and the quantization noise introduced by A/D converter 18. For the most part, the quantization noise is further deteriorated by clock jitter introduced by the clocked A/D converter. The quantization noise is particularly troublesome at high frequencies such as the 800 MHz. Consequently, the LMS machine reacts adversely to the unwanted noise and spikes introduced by the conventional slicer.
Conventional Multilevel Slicers
For example, as analog signal y(t) 101 reaches threshold level 102, as illustrated in
This blind process of not knowing the final destination results in jagged edges, and increases the transition period between sampling periods. Increasing the transition period decreases the size of the sampling period (e.g., eye opening as discussed below), which is used to reconstruct the analog signal in the digital domain. The increased transition period and decreased sampling periods create narrower pulses in the error signal. The error signal is generated by subtracting the input of the slicer from the output of the slicer. Furthermore, the error signal has less energy because of the lower quantization levels used to generate d(t) 116. Since the error signal is also used in the LMS machines, the narrower pulses of the error signal may prevent the LMS machine from converging. This results in overall poor performance. It should also be noted that conventional multilevel slicers include complex circuitry to analyze the multilevel thresholds. In addition, conventional multilevel slicers do not work beyond 3 pulse amplitude modulations (PAM).
A method and an apparatus for slicing an analog signal are described. In one embodiment, the method includes slicing an analog input signal in the analog domain to generate an encoded analog signal, and subtracting the analog input signal from the encoded analog signal to provide an analog error signal.
The present invention will be understood more fully from the detailed description given below and from the accompanying drawings of various embodiments of the invention, which, however, should not be taken to limit the invention to the specific embodiments, but are for explanation and understanding only.
A method and apparatus for slicing an analog signal using an analog encoder, and subtracting the analog signal from the encoded analog signal from the analog encoder to produce an analog error signal without using an analog-to-digital converter (A/D converter) and digital-to-analog converter (D/A converter), are described. Because A/D and D/A converters are not used in the slicing operation, the slicer does not require a clock signal. Because the slicer is based on an analog encoder, there are no sharp edges (e.g., infinite slopes introduced between various reference levels) at the output of the slicer, thus, the resulting error signal, which is obtained by subtracting the output of the slicer from the input of the slicer, is smoother than the conventional slicers that are based on A/D and D/A converters. The error signal is used in the LMS machine, and thus, having a smoother error signal allows the LMS machine to be well behaved (e.g., converge to a better solution). By using the analog encoder in the slicer, tap noise is reduced, and the overall performance of the signal processing system is improved.
In the following description, numerous specific details are set forth, such as specific frequencies, in order to provide a thorough understanding of the present invention. It will be apparent to one skilled in the art, that the present invention may be practiced without these specific details. In other instances, well-known circuit elements, such as amplifiers and multipliers, are not described in detail in order to not unnecessarily obscure the present embodiments.
In one embodiment, joint training the equalizer and the echo canceller is performed by providing independent error signals to two LMS machines, one each corresponding to the equalizer and the echo canceller. In one embodiment, joint training does not involve complex start-up process and does not require special timing between the training of the equalizer and the training of the echo canceller. This allows the equalizer and the echo canceller to operate independent from one another, and allows for an independent reset of the equalizer and the echo canceller. This may yield a stable loop in the analog domain, and ultimately, improve performance of the circuit. LMS machines perform least mean square algorithms, and are used in adaptive filters to find the filter coefficients or tap weights that relate to producing the least mean squares of the error signal (e.g., difference between the desired and the actual signal). Although, the embodiments herein are described as using LMS algorithms to determine the tap weights of the adaptive filters, alternatively, other algorithms known by those of ordinary skill in the art of ordinary skill in the art can be used.
As will be described below, many of the signal impairments removed in the digital domain for the arrangement of
Unlike the duplexing circuit of
The input signal from duplexing circuit 32 is coupled to a low pass filter 40 through the high frequency transformer 34. This may be an ordinary analog low pass filter such as typically used to limit the high frequencies of a signal, which is subsequently digitized. Generally, frequencies higher than those that can be faithfully digitized are removed.
The output of filter 40 is coupled to an amplifier 41, which amplifies an input signal based on a gain control signal. Amplifier 41 may be an ordinary analog amplifier, such as used in digital signal processing section. Amplifier 41 typically controls the gain of the received signal so that the amplified signal falls within a predetermined region of the operating characteristics of the circuits receiving the amplified signal.
The output of amplifier 41 is coupled to an analog delay line 45. Delay line 45 has multiple stages such as stages 45 a-45 n, each of which, in one embodiment, provides equal periods of delay. In one embodiment, each stage has one or more segments, and each segment includes an inductor and a capacitor. This allows for a fractionally or symbol spaced equalizer. For the illustrated embodiment, each stage has a single inductor and a single capacitor. Ideally, delay line 45 is lossless; although as a practical matter, there is some loss associated with each of the stages. For purposes of discussion, each of the stages are consecutively numbered from n=1 to n=N.
In one embodiment, the entire delay line is fabricated from passive elements (inductors and capacitors) without amplification between stages. This reduces the noise that would otherwise occur and build up over the delay line. Ideally, the magnitude at each tap is constant with only the phase of the signal changing.
A signal at a tap from each stage of delay line 45 is coupled to two combining circuits. Specifically in
Multiplier 52 multiplies the output of delay circuit 51 by an error signal e(t) on line 54. The resultant analog signal from multiplier 52 is coupled to an integrator 56. Integrator 56, which may be an ordinary capacitance integrator, performs integration on the analog signal from multiplier 52. In one embodiment, integrator 56 has a time constant measured in microseconds for a received signal in the GHz range. Thus, this integration is relatively long with respect to the period of the received signal. The output of integrator 56 is coupled as one input to multiplier 58, which multiplies it by the signal from tap 50.
There is a delay circuit, two combining circuit, and an integrator for each of the taps of delay line 45. The outputs from the second combining circuits, such as multiplier 58, are all combined in summation unit 60. As discussed below, the output of the summation unit provides the analog output signal, y(t), which is the input signal X(t), which is shown in
The analog error signal on line 54 is generated by slicing the signal y(t) in slicer 61 and then subtracting the resultant signal from the input to slicer 61 in subtractor 62. This results in an error signal, which is used, as described below, to develop the adaptive tap weights forming one input to the second combining circuit (e.g., multiplier 58).
The embodiment of
In one embodiment, the circuitry of
The circuitry of
Y(t)=ΣW n ·X[t−(n−1)T]
where y(t) is the signal at the output of summation unit 60, is the adaptive tap weight associated with tap n, and X(t-nT) is an output at a tap for a stage n, the input function at the taps for each stage n, where T is equal to the time delay of each of the stages. This equation can be expanded as follows:
Y(t)=W 0 X(t)+W 1(t−T)+W 2 X(t−2T)+ . . . W N X(t−NT)
where N+1 is the total number of stages in the delay line. Each term in this equation has a value represented by the output of the second combining means such as multiplier 58. The terms are then summed within summation unit 60 to generate y(t).
The tap weights for the embodiment of
is a constant, e(t) is the error signal on line 54, and τ is the delay provided by the differential delay circuit 51. The integration shown in the above equation is performed by integrator 56.
As may be noted from
In digital signal processing, this loop stabilizing delay is not required. In the digital domain, a value representing an error signal, for instance, can be readily stored and then used as feedback to integrator 56 under the control of a timing signal, and thus, the feedback problem described above does not occur.
An alternate embodiment is shown in
If τ is equal to 2T, then the signal at the tap of stage n+2 provides the same delay as differential delay circuit 51 of
Performance of the Duplexing Circuit
The performance of the duplexing circuits of
Joint Training of Equalizer and Echo Canceller
Unlike conventional phased training, joint training does not require a complex start-up process. Because joint training does not require complex start-up process, no complex timing is necessary during joint training. Furthermore, the performance of AEQ 76 is independent from the performance of AEC 73 because both AEQ 76 and AEC 73 receive independent error signals 79B and 79C, respectively. In one embodiment, when AEQ 76 has not trained properly, meaning error signal 79B is large (e.g., error signal 79B is large, such that signal-to-noise ratio (SNR) is less than 10 dB), the error signal 79B may not affect error signal 79C because the training of AEC 73 is performed independently of the training of AEQ, and with an independent error signal 79C. In another embodiment, when AEC 73 has not trained properly, indicated by error signal 79C being large, error signal 79C does not affect error signal 79B because the training of AEQ 76 is performed independently of the training of AEC 73, and with an independent error signal. As described above, having a large error signal, for either error signal 79B or 79C, may be the result of having an incorrect tap weight. In addition, the tap weights may be incorrect if the loop becomes unstable.
In one embodiment, the output signals of AEQ 76 and AEC 73 are orthogonal with respect to each other. In one embodiment, in order to have the output signals of AEQ 76 and AEC 73 be orthogonal to each other, time constant of the integrator 56 needs to be large to slow down the circuit. The time constant should be larger than the loop delay, such as one or more orders of magnitude larger of than the loop delay. As described above, if this feedback has a long delay, instability can occur since the feedback may cause the signal at the output of integrator 56 to continually rise or fall. Having a large time constant may also reduce the power consumed by the circuit.
It should be noted that joint training permits AEQ 76 and AEC 73 to be reset independent of one another. When the AEQ 76 and AEC 73 are reset, the tap weights of the AEQ 76 and AEC 73 are set to an initial condition, where the tap weights may ultimately diverge to the desired solution. Also, because the far-end transmitter is orthogonal with respect to the near-end transmitter, the AEC 73 correlates the near-end transmitter to the near-end echo to remove the echo from received signal. The existence of the estimated echo signal does not affect the equalization because the AEC 73 is set to correlate the near transmitter to the near-end echo, and is not correlated to the received signal received by the AEQ 76. The time constant of the integrator of the AEC 73 needs to be large, as described above, in order to not affect the received signal (e.g., not integrating the received signal). The echo can be effectively removed from the received signal without affecting the equalization of the received signal.
In one embodiment, the method of joint training is performed by training the AEQ 76 using analog error signal 79B and LMS machine 77, and training AEC 73 using analog error signal 79C and LMS machine 74. The training of AEQ 76 and AEC 73 is performed jointly. It should be noted that the analog error signal 79B of the LMS machine 77 and the analog error signal 79C of the LMS machine 74 are independent of each other.
In one embodiment, error signal 79B may be obtained by slicing an output signal of AEQ 76 to generate a sliced signal, and subtracting the sliced signal of AEQ 76 and the output signal of AEQ 76 to generate the analog error signal 79B of the LMS machine 77. This process is known as blind training.
In another embodiment, error signal 79C may be obtained by receiving an input signal from either duplexing circuit 10 of
It should be noted that the output signal of AEQ 76 is coupled to the input of slicer 61, as well as the input of an A/D converter 49. Furthermore, A/D converter 49 may be coupled to a digital signal processing unit.
Training of AEQ 76 (or AEC 73) includes determining and adjusting the tap weights of a delay line, as described above with respect to
Analog Encoder Based Slicer
In one embodiment, the method of slicing an analog signal is performed by slicing an analog input signal y(t) on line 94 in the analog domain to generate an encoded analog signal d(t) on line 95, and subtracting the analog input signal y(t) on line 94 from the encoded analog signal d(t) on line 95 to provide an analog error signal e(t) on line 96. Thus, the analog signal y(t) on line 94 can be sliced using an analog encoder 91 and subtracted using subtractor 93 from the encoded analog signal d(t) on line 95 to generate the analog error signal e(t) on line 96.
Because the analog signal is sliced in the analog domain, slicer 61 is placed in parallel to A/D converter 49. For example, the analog signal y(t) that is input into slicer 61 on line 94 is also input into A/D converter 49. This permits slicer 61 to operate without a clock signal, reducing the quantization noise introduced by clock jitter, as well as other digital conversion noise. Thus, slicer 61 uses an analog signal with a higher signal-to-noise ratio to generate the error signal e(t) on line 96 to be used as feedback for the LMS machines.
Thus, improved front end processing has been described for a data carrying signal received over a twisted pair. Many of the impairments often removed with a digital signal processing in the prior art are removed in the analog domain. This, as mentioned herein, significantly reduces the performance required of the A/D converter, and thereby provides a more readily realizable and better performing circuit.
In one embodiment, in generating the encoded analog signal d(t), the analog encoder 91 saturates the encoded analog signal d(t) on line 95 at an upper saturation value (e.g., VSAT+ 97) when the analog signal y(t) on line 94 is above an upper threshold value (e.g., ε+ 99). The analog encoder 91 also saturates the encoded analog signal d(t) at a lower saturation value (e.g., VSAT− 98) when the analog signal y(t) on line 94 is below a lower threshold value (e.g., ε− 100). In one embodiment, analog encoder 91 generates the encoded analog signal d(t) on line 95 having a characteristic of a finite slope when the analog signal y(t) on line 94 is between the upper and lower threshold values, ε+ 99 and ε− 100, respectively.
In one embodiment, the upper saturation value (e.g., VSAT+ 97) is greater than the upper threshold value (e.g., ε+ 99) and the lower saturation value (e.g., VSAT− 98) is less than the lower threshold value (e.g., ε− 100). Alternatively, the upper saturation value (e.g., VSAT+ 97) is equal to the upper threshold value (e.g., ε+ 99) and the lower saturation value (e.g., VSAT − 98) is equal to the lower threshold value (e.g., ε− 100).
In one embodiment, analog encoder 91 may be a comparator that compares the analog signal y(t) (e.g., on line 94) to a predetermined set values, namely upper and lower threshold values, ε+ 99 and ε− 100, respectively, to generate, the encoded analog signal d(t) (e.g., on line 95). As described above, subtractor 93, which is coupled to the output of the comparator (analog encoder 91) and the input of the comparator, subtracts the analog signal y(t) (e.g., on line 94) from the encoded analog signal d(t) (e.g., on line 95) to produce the analog error signal e(t) (e.g., on line 96).
The analog error signal e(t) on line 96 may be used as a feedback error signal for the LMS machine. In one embodiment, the analog error signal e(t) on line 96 is analog error signal 79B received by LMS machine 77, which is used in joint training of AEQ 76, as described with respect to
Having described and illustrated the problem with jitter as well as the problem of having jagged edges, it should be noted that due to the jagged edges introduced by the multilevel slicer, as described with respect to
In one embodiment, by employing a two-level slicer instead of a multilevel slicer, the two-level slicer does not introduce jagged edges like the multilevel slicer. As a result, the calculated error signal e(t), derived from subtracting the input from the output of the two-level slicer, may not have narrow pulses like the multilevel slicer, resulting in the correct energy in the sliced signal. Furthermore, the analog error signal having non-narrow pulses and correct energies permits the LMS machine to converge, which may improve the overall performance of the system.
A two-level slicer includes a much simpler circuit than a multilevel slicer, such as the five threshold level slicer described above. Furthermore, the two-level slicer as described herein may function with a multilevel signal having PAM up to 16 PAM and higher.
Note that in one embodiment, the 2-level and multi-level slicers are clockless (i.e., they do not require a clock signal). Because the embodiments do not require a clock signal, there is no complex timing involved. Furthermore, it should be noted that eventually a clock signal may be generated, but the clock signal may be extracted from a healthy signal, a signal with a higher signal-to-noise ratio, as opposed to a weak signal that includes a lot of noise. For example, the clock may be generated after removing the impairments, such as near end and far end crosstalk echo and ISI. Furthermore, because this processing is performed in the analog domain before converting the analog signal into a digital signal there may be no quantization noise, in turn deteriorated further by clock jitter (e.g., introduced by A/D converter 18), reducing the overall signal-to-noise ratio. Because A/D converter 18 receives analog signal having a lower signal-to-noise ratio, A/D converter 18 typically requires only 7 ENOB for subsequent processing, as opposed to the 10 or 11 ENOB required in the prior art, as shown and described with respect to
Thus, improved front end processing has been described for a data carrying signal received over a twisted pair. Many of the impairments often removed with digital signal processing are removed in the analog domain. This significantly reduces the performance required of the A/D converter, as described above, and thereby provides a more readily realizable and better performing circuit.
Although present embodiments have been described with reference to specific embodiments, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the present embodiments. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4074086||Sep 7, 1976||Feb 14, 1978||Bell Telephone Laboratories, Incorporated||Joint adaptive echo canceller and equalizer for two-wire full-duplex data transmission|
|US4439639 *||Mar 19, 1982||Mar 27, 1984||Northern Telecom Limited||Digital circuit and method for the detection of call progress tones in telephone systems|
|US4577329 *||Oct 6, 1983||Mar 18, 1986||Telecommunications Radioelectriques Et Telephoniques T.R.T.||Self-adaptive equalizer for baseband data signals|
|US4918685 *||Jul 15, 1988||Apr 17, 1990||At&T Bell Laboratories||Transceiver arrangement for full-duplex data transmission comprising an echo canceller and provisions for testing the arrangement|
|US5347539||Jan 15, 1993||Sep 13, 1994||Codex Corporation||High speed two wire modem|
|US5744385||Mar 21, 1997||Apr 28, 1998||Plato Labs, Inc.||Compensation technique for parasitic capacitance|
|US5880634||Mar 21, 1997||Mar 9, 1999||Plato Labs, Inc.||Wide band-width operational amplifier|
|US5936445||Mar 21, 1997||Aug 10, 1999||Plato Labs, Inc.||PLL-based differential tuner circuit|
|US6028479||Jan 7, 1998||Feb 22, 2000||Plato Labs, Inc.||Low voltage transmission line driver|
|US6069505||Mar 19, 1998||May 30, 2000||Plato Labs, Inc.||Digitally controlled tuner circuit|
|US6169764 *||Mar 19, 1998||Jan 2, 2001||Plato Labs, Inc.||Analog adaptive line equalizer|
|US6256157 *||May 15, 1998||Jul 3, 2001||International Business Machines Corporation||Method and apparatus for removing noise spikes|
|US6266367||May 28, 1998||Jul 24, 2001||3Com Corporation||Combined echo canceller and time domain equalizer|
|US6600775||Aug 8, 2002||Jul 29, 2003||Echelon Corporation||Adaptive reference pattern for spread spectrum detection claims|
|US6778599||Mar 9, 2000||Aug 17, 2004||Tioga Technologies||Digital transceiver with multi-rate processing|
|US6795494||May 12, 2000||Sep 21, 2004||National Semiconductor Corporation||Receiver architecture using mixed analog and digital signal processing and method of operation|
|US6909742||Feb 27, 2004||Jun 21, 2005||Finisar Corporation||Method and apparatus for reducing interference in a data stream using autocorrelation derived equalization|
|US6928110||May 22, 2001||Aug 9, 2005||Panasonic Communications Co., Ltd.||Adaptive equalizer training circuit, modem apparatus and communication apparatus|
|US6958549||Apr 26, 2004||Oct 25, 2005||Honda Motor Co., Ltd.||Hybrid type driving apparatus|
|US6985549||Jan 17, 2001||Jan 10, 2006||Ati Research, Inc.||Blind cost criterion timing recovery|
|US6990163 *||Sep 6, 2001||Jan 24, 2006||Lsi Logic Corporation||Apparatus and method for acquiring phase lock timing recovery in a partial response maximum likelihood (PRML) channel|
|US7471904 *||Jul 25, 2003||Dec 30, 2008||Alcatel-Lucent Usa Inc.||Method and apparatus for electronic equalization in optical communication systems|
|US7483479 *||Sep 16, 2004||Jan 27, 2009||Keyeye Communications||Scaled signal processing elements for reduced filter tap noise|
|US7535956 *||Jul 25, 2005||May 19, 2009||Fujitsu Limited||Adaptive equalizer circuit|
|US7561619||Dec 19, 2003||Jul 14, 2009||Intel Corporation||Feedback filter|
|US7839758||Sep 23, 2008||Nov 23, 2010||Net Logic Microsystems, Inc.||Analog echo canceller with interpolating output|
|US7843859||Sep 23, 2008||Nov 30, 2010||Netlogic Microsystems, Inc.||Analog echo canceller with filter banks|
|US20030072363||Oct 15, 2002||Apr 17, 2003||Mcdonald James Douglas||Adaptive equalization of digital modulating signal recovered from amplitude-modulated signal subject to multipath|
|US20030235245||Dec 18, 2002||Dec 25, 2003||Erdogan Alper Tunga||Method and system for computing pre-equalizer coefficients|
|US20050019042||Jul 25, 2003||Jan 27, 2005||Noriaki Kaneda||Method and apparatus for electronic equalization in optical communication systems|
|US20050094583||Nov 3, 2003||May 5, 2005||Gupta Sandeep K.||Full duplex transceiver|
|US20060262686 *||Sep 7, 2005||Nov 23, 2006||Sheng-Hung Wu||Digital data slicing circuit and slicing method|
|US20080080598||Feb 16, 2007||Apr 3, 2008||Shaomin Samuel Mo||Method and apparatus for processing communication using different modulation schemes|
|US20080224793||Mar 15, 2007||Sep 18, 2008||Babanezhad Joseph N||Joint phased training of equalizer and echo canceller|
|US20100142604||Dec 9, 2008||Jun 10, 2010||Yehuda Azenkot||Method and apparatus of frequency domain echo canceller|
|EP0601636A1||Dec 1, 1993||Jun 15, 1994||Philips Electronics N.V.||X-ray apparatus comprising wheels provided with cable deflectors|
|WO2008112328A1||Mar 17, 2008||Sep 18, 2008||Plato Networks Inc||Training of equalizer and echo canceller|
|1||"IEEE 802 10GBASE-T Tutorial", (Nov. 2003), 1-56.|
|2||John G. Proakis "Digital Communications" fourth edition 2000.|
|3||PCT International Preliminary Report on Patentabilityfor PCT International Appln. No. US2008/003625, mailed Sep. 15, 2009 (9 pages).|
|4||PCT, International Search Report and Written Opinionfor PCT International Appln. No. US2008/003625, mailed Aug. 5, 2008 (17 pages).|
|5||*||Sedra/Smith "Microelectronic circuits" Fourth Edition 1998. (pp. 14-15).|
|6||Smolka, Georg J., "ALALOG CMOS Circuits for ISDN", ISCAS '88, Siemens AG, Components Group, 8000 Munich, West Germany, (1988), 1927-1930.|
|7||USPTO Office Action dated Jun. 14, 2010, in the patent application entitled "Joint Phased Training of Equalizer and Echo Canceller", with U.S. Appl. No. 11/724,816, filed Mar. 15, 2007.|
|8||USPTO Office Action dated Mar. 17, 2010, in the patent application entitled "New Least Mean Square (LMS) Engine for Multilevel Signal", with U.S. Appl. No. 11/724,817, filed Mar. 15, 2007.|
|9||USPTO Office Action dated Nov. 17, 2010, in the patent application entitled "Joint Phased Training of Equalizer and Echo Canceller", with U.S. Appl. No. 11/724,816, filed Mar. 15, 2007.|
|10||USPTO Office Action dated Nov. 23, 2009, in the patent application entitled "Joint Phased Training of Equalizer and Echo Canceller", with U.S. Appl. No. 11/724,816, filed Mar. 15, 2007.|
|11||USPTO Office Action dated Oct. 20, 2010, in the patent application entitled "New Least Mean Square (LMS) Engine for Multilevel Signal", with U.S. Appl. No. 11/724,817, filed Mar. 15, 2007.|
|12||Widrow, B. , et al., "Adaptive Antenna Systems", Proceedings of the IEEE, Vo. 55, No. 12, Dec. 1967, 2143-2159.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8619897||Dec 9, 2008||Dec 31, 2013||Netlogic Microsystems, Inc.||Method and apparatus of frequency domain echo canceller|
|US8723488 *||Dec 20, 2011||May 13, 2014||The Powerwise Group, Inc.||IGBT/FET-based energy savings device for reducing a predetermined amount of voltage using pulse width modulation|
|US20100142604 *||Dec 9, 2008||Jun 10, 2010||Yehuda Azenkot||Method and apparatus of frequency domain echo canceller|
|US20120086415 *||Dec 20, 2011||Apr 12, 2012||The Powerwise Group, Inc.||IGBT/FET-Based Energy Savings Device for Reducing a Predetermined Amount of Voltage Using Pulse Width Modulation|
|U.S. Classification||375/229, 375/233, 333/18, 375/232, 375/234, 333/28.00R|
|Cooperative Classification||H04L25/03885, H04L2025/03617, H04L25/03057, H04B3/23|
|May 21, 2007||AS||Assignment|
Owner name: PLATO NETWORKS, INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BABANEZHAD, JOSEPH N.;REEL/FRAME:019344/0928
Effective date: 20070503
|Sep 29, 2010||AS||Assignment|
Owner name: NETLOGIC MICROSYSTEMS, INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PLATO NETWORKS, INC.;REEL/FRAME:025051/0940
Effective date: 20100921
|Apr 16, 2015||AS||Assignment|
Owner name: BROADCOM CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NETLOGIC I LLC;REEL/FRAME:035443/0763
Effective date: 20150327
Owner name: NETLOGIC I LLC, DELAWARE
Free format text: CHANGE OF NAME;ASSIGNOR:NETLOGIC MICROSYSTEMS, INC.;REEL/FRAME:035443/0824
Effective date: 20130123
|Jul 2, 2015||REMI||Maintenance fee reminder mailed|